I would like to check with you my following questions:
For resting state connectivity analysis, when I indicate an ROI, and then run “connectivity\Coherence”, does the software calculate the coherence between indicated ROI and all cortex regions?
What is different between N*N and 1*N coherence in term of calculation?
What about if I indicate 2 ROIs? Only coherence between these 2 regions is calculated? However, I got out of memory problem to run it, what can I do?
The menus [1xN] estimate the connectivity between one seed (any signal: sensor, source, scout) and all the other values in the same file (all the sensors or all the other sources).
The menus [NxN] estimate the connectivity between all the pairs of signals within a file (all the sensors, all the sources or all the scouts).
Additionally, you have menus [AxB] in the Process2 tab, with which you get additional flexibility to compute connectivity metrics between the pairs of signals you want from your database (sensors x sources, sensors x scouts, etc.)
If you select a source file in Process1, then “Coherence [1xN]”, select [B]one scout[/B] in the scout list, and select the option “When to apply the scout function: [B]After[/B]” a it will:
Estimate the coherence between all the sources of the scouts (let’s say your scout contains Ms sources) and all the sources in the source space that represents your brain (let’s say you have Ns sources). This gives you a [Ms x Ns x freq] coherence matrix.
Group the all the Ms values into one using the scout function you select in the options of the process. In the end you get a matrix [1 x Ns x freq]
If you do the same but with the option “When to apply the scout function: [B]Before[/B]”, it does the following
Average the activity of the Ms signals of the scout into one signal
Estimate the coherence between this averaged scout signal and all the brain. This ouput size is the same: matrix [1 x Ns x freq]
The first option is (after) usually has higher chances to extract the highest connectivity patterns you may observe between to pairs of sources. This is the recommended approach.
In the processes [1xN], if you select two scouts, you will obtain a [2xNs] connectivity matrix, which you will not be able to display. If you want to study the interactions between two scouts, you should use the [NxN] process.
All those processes are very memory demanding, especially if you are trying to do a full cortex analysis (remember you have 15000 signals at least…)
Alternatively: work only with regions of interest, or get more memory (a LOT more)